do not skip span for newline textnodes otherwise compute_new_comment_positions() will return bad results for pandoc texts.
from django.utils.translation.trans_real import translation
def translate_to(value, language_code):
t = translation(language_code)
return getattr(t, 'gettext')(value)